Mailman 3 November 2013 - scikit-image

Re: Advice on basic operations: zoom, crop and splice
by Stéfan van der Walt Nov. 23, 2013

Nov. 23, 2013

On Sat, Nov 23, 2013 at 8:16 AM, Chintak Sheth <chintaksheth(a)gmail.com> wrote: > An interesting thread regarding interactive image > manipulation in iPython notebook came up lately which might interest you: > https://groups.google.com/forum/#!topic/scikit-image/vrPzanJ42rs. I spoke to Brian Granger about the possibility of building the viewer described above in IPython. Here is his response: """ Definitely possible - shouldn't be too bad. Here is a rough sketch: * You can define a Python object with traitlets. * You get a JavaScript object that has the same exact attributes as the Python object - we keep those attributes completely synchronized on the two sides automatically. When the JavaScript side (through UI control changes) changes the attributes we sync the python side and all of the traitlets callbacks will fire. When the Python side makes changes, the JS side is updated and the UI will refresh. The time scale for all of this is super fast - at least running on localhost. * You can then create arbitrary "view" code for the JavaScript side. You get a div to put the UI in and can do whatever you want. * The JavaScript and Python objects also have methods for sending arbitrary JSON messages at any time, so you can stream data, etc. To get familiar with how it all works, have a look at the notebooks in this branch: https://github.com/jdfreder/ipython/tree/widget-msg/examples/widgets They show how to develop new widgets from scratch as well as how all of the existing ones can be used. """ I am really keen to try this out next week, although I may have to up my Javascript-foo a bit. Stéfan

1 0

Re: Advice on basic operations: zoom, crop and splice
by Chintak Sheth Nov. 23, 2013

Nov. 23, 2013

Hi Adam On Nov 23, 2013 12:33 AM, "Adam Hughes" <hughesadam87(a)gmail.com> wrote: > > 3. If possible, removing the ROI from the original image, and splicing the original image back together. If this is possible, that would be amazing. This would allow us to effectively cut out regions of our images that are obviously contaminates. Yes, this is possible. It is also known as Image Inpainting. There are two algorithm implementations for the same, although they haven't yet been merged. Fast Marching Algorithm: (Fast, introduce some blurriness) Here is the dev branch: https://github.com/chintak/scikit-image/tree/inpaint PR for it: https://github.com/scikit-image/scikit-image/pull/663 Exemplar based Inpainting: (Slower, very accurate for repeating patterns, reconstructs using sample image patches from the image) Dev branch: https://github.com/chintak/scikit-image/tree/inpaint_exemplar PR: https://github.com/scikit-image/scikit-image/pull/706 For the first one, a skimage.viewer plugin is also there: ``viewer_examples/plugins/demo_inpaint.py`` Regarding your first 2 points, Josh shed some light on how you can do this using `skim age.viewer`. An interesting thread regarding interactive image manipulation in iPython notebook came up lately which might interest you: https://groups.google.com/forum/#!topic/scikit-image/vrPzanJ42rs. I'm not sure if you can select an ROI in the image, but that'd be worth checking out. If we can't directly select a rectangular region then probably defining 4 sliders in the widget: left, right, top, bottom, and moving them around to select the ROI. Chintak

1 0

Re: Advice on basic operations: zoom, crop and splice
by Josh Warner Nov. 22, 2013

Nov. 22, 2013

The backend functionality for all of this exists. I believe all of this can be accomplished with masks and NumPy array slicing. The problem is that generally speaking, these are not interactive or GUI level operations. For example, if you want a zoomed-in ROI you can slice that region out and then `plt.imshow(sliced_img)`, but this requires manually entering the ROI corners in the slicing operation. `skimage.viewer` includes a method to select and save rectangular regions interactively, and could probably be extended to allow clipping them out as well. However, the viewer requires PyQt or PySide and cannot be embedded in an iPython notebook. Are you specifically looking for iPython notebook capable solutions? Or would a tweaked Viewer plugin be sufficient? On Friday, November 22, 2013 1:03:04 PM UTC-6, Adam Hughes wrote: > > Hi everyone, > > I'm used to working with big images of lots of particles. In the > notebook, I'd like to be able to look at a full image, and then at a zoomed > in region of interest. A few basic questions come to mind: > > Is there a zoom/crop function or preffered approach to basic > manipulations of zooming and cropping, or would I have to do this at the > numpy or matplotlib level? I saw that there's a rectangle function that > probably would be useful here. Does anyone have any examples or personal > code built for doing some of these common manipulations? Ideally, I want > to take the fastest approach to: > > 1. Selecting a rectangular region of interest (ROI). > 2. Cropping or zooming in on this region, and storing the ROI as its own > array/image. > 3. If possible, removing the ROI from the original image, and splicing > the original image back together. If this is possible, that would be > amazing. This would allow us to effectively cut out regions of our images > that are obviously contaminates. > > Thanks. > > >

1 0

Advice on basic operations: zoom, crop and splice
by Adam Hughes Nov. 22, 2013

Nov. 22, 2013

Hi everyone, I'm used to working with big images of lots of particles. In the notebook, I'd like to be able to look at a full image, and then at a zoomed in region of interest. A few basic questions come to mind: Is there a zoom/crop function or preffered approach to basic manipulations of zooming and cropping, or would I have to do this at the numpy or matplotlib level? I saw that there's a rectangle function that probably would be useful here. Does anyone have any examples or personal code built for doing some of these common manipulations? Ideally, I want to take the fastest approach to: 1. Selecting a rectangular region of interest (ROI). 2. Cropping or zooming in on this region, and storing the ROI as its own array/image. 3. If possible, removing the ROI from the original image, and splicing the original image back together. If this is possible, that would be amazing. This would allow us to effectively cut out regions of our images that are obviously contaminates. Thanks.

1 0

Re: Fwd: Fun with skimage
by Josh Warner Nov. 21, 2013

Nov. 21, 2013

If we want popular exposure, stuff like - Small snippets that can create Instagram-like filters in a few lines. - Fiddling with transforms to yield some interesting effects. - Perhaps combination face detection + transforms? would probably track quite well. On Wednesday, November 20, 2013 10:28:57 PM UTC-6, Juan Nunez-Iglesias wrote: > > On Thu, Nov 21, 2013 at 12:30 PM, Marianne Corvellec < > marianne....(a)ens-lyon.org <javascript:>> wrote: > >> Maybe we could add a "Fun" subsection to the draft structure under >> "Purpose" (scikit-image paper)... :p >> >> I think the matplotlib people mentioned the most popular part of the >> matplotlib gallery was the XKCD-stylification of plots... I quite like >> that. >> > > +1! Maybe that embedded face detection example could go in there. =) > >

1 0

Re: Advice on improving a analysis on images of nanoparticles
by Guillaume Gay Nov. 21, 2013

Nov. 21, 2013

Hi, This: http://damcb.com/posts/segmenting-nuclei-with-skimage.html links to a notebook where I expose what I'm doing for the segmentation. It won't be too much work to turn it into an example if needed. Also feel free to comment on the post... As you'll see I kind of go around the 3D part by clustering strategies, more that by doing real 3D stuff... Cheers. Guillaume PS: I'll try to have syntax highlighting in nikola soon, so that this code is more readable. On 20/11/2013 13:55, Juan Nunez-Iglesias wrote: > @Guillaume, please contribute the example ASAP because I'm struggling > with the exact same problem! =D In particular, the nuclei example from > pythonvision is not ideal because they gloss over a lot of parameter > tuning that is much easier to deal with in 2D than 3D. I'll be happy > to hear about your experience! > > Juan. > > > On Wed, Nov 20, 2013 at 9:50 PM, Brickle Macho <bricklemacho(a)gmail.com > <mailto:bricklemacho@gmail.com>> wrote: > > I can always adapt this example > (http://pythonvision.org/basic-tutorial) which uses pymorph and > mahotaos to counting nuceli and segments the image as a gallery > example for skimage (obviously acknowledging the original > tutorial). Or is it bad form to copy/use another example. > > Michael. > -- > > > > > On 20/11/2013 2:13 pm, Stéfan van der Walt wrote: >> >> On 20 Nov 2013 07:33, "Juan Nunez-Iglesias" <jni.soma(a)gmail.com >> <mailto:jni.soma@gmail.com>> wrote: >> > >> > I'm guessing you are applying label() directly to your image, >> which is not the right way to use it. >> >> Looks like isolating objects is a common problem. Time for >> another gallery example? We already have the coins, but perhaps >> something more realistic with some biological data. >> >> Stéfan >> >> -- >> You received this message because you are subscribed to the >> Google Groups "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, >> send an email to scikit-image+unsubscribe(a)googlegroups.com >> <mailto:scikit-image+unsubscribe@googlegroups.com>. >> For more options, visit https://groups.google.com/groups/opt_out. > > -- > You received this message because you are subscribed to the Google > Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, > send an email to scikit-image+unsubscribe(a)googlegroups.com > <mailto:scikit-image%2Bunsubscribe@googlegroups.com>. > For more options, visit https://groups.google.com/groups/opt_out. > > > -- > You received this message because you are subscribed to the Google > Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send > an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out.

2 1

Re: scikit-image paper
by Stéfan van der Walt Nov. 21, 2013

Nov. 21, 2013

On Thu, Nov 21, 2013 at 6:39 AM, Tony Yu <tsyu80(a)gmail.com> wrote: > Unfortunately, it's a no-go on mentioning the specific companies we've > worked with, but its OK to say something like "Enthought, Inc uses > scikit-image extensively in their consulting projects related to geophysics > and microscopy". That's great, thank you! Could I use [1] as a reference? Stéfan [1] Personal communications with Tony S Yu, <insert title> [or insert other name here]

1 0

Re: scikit-image paper
by Tony Yu Nov. 21, 2013

Nov. 21, 2013

On Mon, Nov 18, 2013 at 10:49 PM, Tony Yu <tsyu80(a)gmail.com> wrote: > > > > On Mon, Nov 18, 2013 at 8:50 PM, Marianne Corvellec < > marianne.corvellec(a)ens-lyon.org> wrote: > >> >> <snip> > I believe Tony S Yu would have lots to share as well. :) >> > > Thanks for pinging me. > > Unfortunately, the current project I'm working on isn't public (at least > when I last checked). There's another enthought project that is public: > I'll ask around to see if it's OK to make a direct reference to the project. > Unfortunately, it's a no-go on mentioning the specific companies we've worked with, but its OK to say something like "Enthought, Inc uses scikit-image extensively in their consulting projects related to geophysics and microscopy". -T

1 0

Re: Fwd: Fun with skimage
by François Boulogne Nov. 21, 2013

Nov. 21, 2013

Le 20/11/2013 05:16, Stéfan van der Walt a écrit : > How cool is this? > > the coffee cup! :) -- François Boulogne. http://www.sciunto.org GPG fingerprint: 25F6 C971 4875 A6C1 EDD1 75C8 1AA7 216E 32D5 F22F

3 2

Re: Advice on improving a analysis on images of nanoparticles
by Juan Nunez-Iglesias Nov. 21, 2013

Nov. 21, 2013

Hi Adam, more responses inline below. =) On Thu, Nov 21, 2013 at 5:04 AM, Adam Hughes <hughesadam87(a)gmail.com> wrote: > > I tried this but to share a link, it asks for the emails of the > recipients. You have used Dropbox to host a publicly accessible link? If > so, I will certainly start doing this, thanks. > The web interface is a bit wonky that way, but it's certainly possible, it's called something like "copy share link". Then you get a URL that has a random token embedded in it, which anyone with the link can open. In OSX, you can just right-click > share dropbox link to have the link copied to your clipboard. That's how I got this: https://www.dropbox.com/s/ic04w1d0x98gnop/suspension%20bridge.jpg =) (You can add a "?dl=1" to the end to go straight to the file.) Yes, that is the goal. We had done a similar process ImageJ, but did > thersholding manually. I will read into the adaptive threshold a bit more. > We had hoped that some of these corrections, such as histogram > equilization, would make the automatic threshold more likely to give > correct results. > adaptive is automatic, but adjusts the threshold individually for each pixel based on surrounding pixels. This should help e.g. for your image warped_f2_b1, where the background is not uniform and no single threshold may work for your entire image. Hmm I see. I will still try it out, but thanks for the heads up. I'll > feel better now if it doesn't work well. > Great strategy. =D We do have an a-prior knowledge actually. What I've been doing already is > putting a lower limit on particle size, with anything under it being noise. > After doing particle counts and binning the data, we fit it with a > guassian, and optionally scale the data so that the guassian is centered > around the mean partitcle diameter (which believe we know to about 3nm > based on TEM imaging and indirect spectroscopic techniques). Based on the > size distribution, we try to further bin the data into small > (dimers/trimers) and large aggregates. For all the particles that are > large enough to be considered an aggregate, we *assume *that they fill a > half-sphere volume, and then we infer the true particle due to these > aggregates. It's pretty ad-hoc, but we certainly apply some knowledge of > the expected particle size distributions. I realize watershedding won't > split up huge clumps, but maybe could assist in the dimers and trimers? In > any case, even if it doesn't significantly enhance our results, it would > still be helpful to explore that option and I'll try it out. > That's the strategy I would suggest, but my point is that in some images, such as your last one, you have more of a carpet of particles, and no single particle will separate, so you will need knowledge from different images. If you have that, no problem! =) If you want to publish the result of your explorations as an IPython notebook, we won't stop you. =D Juan.

1 0